Objective: In an 8-week nonpharmacological pain intervention trial among older adults with lower extremity osteoarthritis (OA), we aimed to examine: a) the baseline frailty level of the participants; b) whether such intervention is more beneficial for baseline frailer older adults than for their counterparts with less frailty; and c) whether the intervention could also alter frailty. Methods: Participants were randomly assigned to either chair yoga (CY) or health education program (HEP) groups and attended twice-weekly 45-minute CY or HEP sessions for 8 weeks. Following a standard procedure, 82 variables were used to construct a frailty index (FI, 0–1). Primary outcomes were: Western Ontario and McMaster Universities (WOMAC) pain and pain interference. Linear mixed-effects models were used to evaluate the modifying effect of baseline frailty on the intervention effect of CY on primary outcomes. Similar models were used to evaluate the effect of CY on frailty. Results: A total of 112 participants (n = 63 CY, n = 49 HEP; 75.3 [SD = 7.5] years) with 85 females (75.9%) were included. The mean values of baseline FI for the CY and HEP groups were similar (0.428 [0.05] and 0.433 [0.05], P = 0.355). Each 0.01 increment in baseline FI was associated with higher WOMAC pain (beta = 0.28, P < 0.001) and pain interference (beta = 0.51, P < 0.001). There was a significant interaction effect between intervention, time, and baseline FI (P = 0.020 for WOMAC pain; P = 0.010 for pain interference), indicating that participants with higher level of baseline FI had greater declines in WOMAC pain and pain interference. There was no significantly greater decline in FI for the CY group compared to the HEP group (between-group difference − 0.01; P = 0.509) and there were no significant trend changes in FI (P for interaction = 0.605). Conclusions: Frailty modifies the intervention effect of CY on pain among older adults with lower extremity OA, underscoring the importance of assessing frailty to improve the management of pain in this population.